Comprehensive quantification of ceramide species in human stratum corneum.

One of the key challenges in lipidomics is to quantify lipidomes of interest, as it is practically impossible to collect all authentic materials covering the targeted lipidomes. For diverse ceramides (CER) in human stratum corneum (SC) that play important physicochemical roles in the skin, we developed a novel method for quantification of the overall CER species by improving our previously reported profiling technique using normal-phase liquid chromatography-electrospray ionization-mass spectrometry (NPLC-ESI-MS). The use of simultaneous selected ion monitoring measurement of as many as 182 kinds of molecular-related ions enables the highly sensitive detection of the overall CER species, as they can be analyzed in only one SC-stripped tape as small as 5 mm x 10 mm. To comprehensively quantify CERs, including those not available as authentic species, we designed a procedure to estimate their levels using relative responses of representative authentic species covering the species targeted, considering the systematic error based on intra-/inter-day analyses. The CER levels obtained by this method were comparable to those determined by conventional thin-layer chromatography (TLC), which guarantees the validity of this method. This method opens lipidomics approaches for CERs in the SC.

novo pathways, such as alterations of some species of the targeted lipidomes into others. Similar to common quantitative analyses using direct MS or LC-MS that need authentic standards together with internal ones, quantitative lipidomics may also require a standard set of all lipid species targeted, because each lipid species has a different molar response in MS detection except for special cases. However, it is practically impossible to obtain all authentic species, including not only different even straight carbon chains but also odd and/or branched ones. To overcome such limitations, lipidomics researchers have contrived a novel procedure to quantify lipidomes comprehensively. Han and Gross ( 3 ) proposed a quantitative method for analyzing triglyceride molecular species using direct MS, which involves a procedure to calculate response factors for species that are not available as authentic materials, based on differences of the responses in the numbers of total carbons and double bonds among the species. For quantitative determination of phospholipid molecular species, Koivusalo et al. ( 4 ) utilized the following parame-Abstract One of the key challenges in lipidomics is to quantify lipidomes of interest, as it is practically impossible to collect all authentic materials covering the targeted lipidomes. For diverse ceramides (CER) in human stratum corneum (SC) that play important physicochemical roles in the skin, we developed a novel method for quantifi cation of the overall CER species by improving our previously reported profi ling technique using normal-phase liquid chromatography-electrospray ionization-mass spectrometry (NPLC-ESI-MS). The use of simultaneous selected ion monitoring measurement of as many as 182 kinds of molecular-related ions enables the highly sensitive detection of the overall CER species, as they can be analyzed in only one SC-stripped tape as small as 5 mm × 10 mm. To comprehensively quantify CERs, including those not available as authentic species, we designed a procedure to estimate their levels using relative responses of representative authentic species covering the species targeted, considering the systematic error based on intra-/inter-day analyses. The CER levels obtained by this method were comparable to those determined by conventional thin-layer chromatography (TLC), which guarantees the validity of this method. This method opens lipidomics approaches for CERs in the SC. - Supplementary key words high sensitivity • normal-phase liquid chromatography-electrospray ionization-mass spectrometry • relative response • simultaneous selected ion monitoring • skin To perform lipidomics ( 1 ), advanced analytical technologies, such as direct MS and LC-MS, are preferably used (2)(3)(4)(5)(6)(7)(8)(9)(10). To characterize differences in lipidomes among biological samples of interest, the relative intensities or areas given from mass spectrometers are only helpful as semi-quantitative values ( 11,12 ). Precise quantitative values of lipidomes of interest would also be desirable, especially in the case of evaluations of lipid dynamics in de MS has been employed for the analysis of CERs in human SC as follows: reversed-phase LC (RPLC)-ESI-MS ( 22,(30)(31)(32); ESI-MS ( 33,34 ); RPLC-thermospray-MS ( 35 ); and NPLC-APCI-MS (36)(37)(38)(39). However, those methods were not validated as quantitative methods for overall CER species in the SC, although some studies reported quantitative values based on unrealistic assumptions. Further, those methods do not have suffi cient detection sensitivity to analyze CERs in smaller SC specimens.
The aim of this study was to develop a novel method for the comprehensive and highly sensitive quantifi cation of CER species in the SC by improving the profi ling method using a recently developed NPLC-ESI-MS technique ( 24 ). We successfully developed such a method using highly sensitive selected ion monitoring (SIM) measurement of as many as 182 kinds of molecular-related ions derived from overall CER species in NPLC-ESI-MS combined with a calculation procedure using relative responses of representative authentic CERs to quantify the overall CER species, including those not available as authentic materials. Herein, we demonstrate that this method can comprehensively quantify CER levels with high sensitivity and that it is useful for practical analyses.

MATERIALS AND METHODS
Nomenclature CER classes are termed according to previous reports ( 21,24,40 ). Briefl y, based on the terminology regarding constitutional fatty acids and sphingoids, the 11

Chemicals
Methanol, 2-propanol, chloroform and n-hexane, of HPLC grade from Kanto Reagents (Tokyo, Japan) were used to prepare SC lipid samples and mobile phases for HPLC. Ultra-pure water prepared using a Milli-Q purifi cation system (Millipore, Bedford, MA) was used in all procedures. All other chemicals were of the highest analytical grade commercially available. Five authentic materials commercially available (purity > 98%), Ceramide ters: (1) molar response factors among the different acyl chain lengths; (2) molar response factors among different degrees of unsaturation; and (3) molar response factors among the different head group classes.
In human cells and tissues, there are commonly three ceramide (CER) classes [combinations of nonhydroxy fatty acids and 4-sphingenines (sometimes referred to as sphingosines); nonhydroxy fatty acids and sphinganines (dihydrosphingosines); and ␣ -hydroxy fatty acids and 4-sphingenines] that contain combinations of C 18 -containing 4-sphingenines or sphinganines and even carbons-containing nonhydroxy or ␣ -hydroxy fatty acids (13)(14)(15). CERs work as bioactive lipids relevant to signal transduction and cell regulation ( 16,17 ). On the other hand, human stratum corneum (SC) is known to contain even more complex CERs which play important physicochemical roles in the barrier and water-holding functions of the skin as structural lipids ( 18,19 ). Thus, it has been revealed that there are seven other CER classes [combinations of nonhydroxy fatty acids and 4-hydroxysphinganines (sometimes referred to as phytosphingosines); nonhydroxy fatty acids and 6-hydroxy-4-sphingenines; ␣ -hydroxy fatty acids and 4-hydroxysphinganines; ␣ -hydroxy fatty acids and 6-hydroxy-4-sphingenines; ester-linked -hydroxy fatty acids and 4-sphingenines; ester-linked -hydroxy fatty acids and 4-hydroxysphinganines; and ester-linked -hydroxy fatty acids and 6-hydroxy-4-sphingenines] in addition to the three common CER classes and the two covalently bound CER classes in the SC (20)(21)(22)(23). However, the structures of such diverse CER species have not been completely characterized. Therefore, we performed structural analyses of those CER species in the SC using a normal-phase-LC-electrospray ionization-MS (NPLC-ESI-MS) technique, where 342 CER species were identifi ed, a new CER class (the eleventh class, combinations of ␣ -hydroxy fatty acid and sphinganines) was discovered, and many kinds of CERs with long chain-containing sphingoids were detected ( 24 ). Three-dimensional multi-mass chromatograms of the overall CER species in NPLC-ESI-MS were useful for CER profi ling among different SC specimens but not for quantifi cation ( 24 ). We further considered the necessity of a quantitative method for the overall CER species to study the statistical relationships between the CER lipidomes and the various types of SC and/or to clarify the de novo biosynthesis of long chain-containing CER species.
Regarding quantifi cation of CERs in the SC, thin-layer chromatography (TLC) has been used conventionally (25)(26)(27). Although TLC can analyze levels of CER classes and remains a quite powerful tool, it cannot be applied to the quantifi cation of individual CER species in the SC. There are other methods that can determine levels of individual CER species, such as HPLC-UV ( 28 ) and GC-MS ( 29 ) with a derivatizing technique, but the CER species targeted for quantifi cation are restricted to only a few species. Over the last decade, many MS methods using electrospray ionization (ESI) or atmospheric pressure chemical ionization (APCI) have been developed and applied to the analysis of CERs in human cells and tissues. Accompanied by such methodological advances, the technology to use kinds of molecular-related ions that have different m/z values and different retention times were simultaneously monitored by means of four monitoring channels and a time program designed in the mass spectrometer (see supplementary Table I). In quantifying levels of CERs in the three specimens each collected from the inner forearm and cheek SC of the subject tested, each crude extracted lipid sample was analyzed in triplicate. The CER levels in the inner forearm and the cheek were presented as the average values of the three specimens.

Calculation of relative responses of authentic CERs
The mixed solutions including both the authentic materials over a concentration range between 10 and 100 nM (10, 20, 50, and 100 nM corresponding to 100, 200, 500, and 1000 fmol in 10 l injection volumes, respectively) and the internal standard (CER[N(17)S(18)]) at a concentration of 50 nM (500 fmol in a 10 l injection volume) were subjected to NPLC-ESI-MS and their calibration curves were then produced. Individual relative responses were calculated as ratios of slopes of calibration curves obtained from targeted authentic materials to that obtained from the internal standard.

Evaluation of inter-day and intra-day reproducibility
The 11 authentic materials belonging to different CER classes were used to clarify the infl uence of different hydrophilic structures on the relative responses. To evaluate the effect of different hydrophobic structures (chain lengths) within the same class on the relative responses, 12 authentic materials all belonging to  (20)], were used. Intra-day reproducibility was obtained from six consecutive analyses on the same day while inter-day reproducibility was obtained from six analyses on separate days over two months.

Calculation of recovery percentage
To estimate quantitative recovery percentage, the 11 authentic materials belonging to different CER classes and the internal standard were spiked in six replicates of crude lipid solutions extracted from the stripped SC to a fi nal concentration of 50 nM each, followed by NPLC-ESI-MS analysis. Recovery percentage was calculated by comparing the determined levels of the 11 authentic CERs in spiked SC lipid samples with those in nonspiked SC lipid samples.

Materials
The protocol was approved by the Ethical Committee of the Kao Corporation of Japan based on the Recommendations from the Declaration of Helsinki. Procedures for the collection of SC specimens with a tape-stripping method and for pretreatment of the SC-stripped tapes were performed with a slight modifi cation of a previously reported method ( 24 ). Tape-stripped SC specimens were collected from three different regions each from the inner forearm and from the cheek of a healthy 35-year old Japanese male, who was enrolled in the study with informed consent. A polyphenylene sulfi de fi lm tape (Nichiban, Tokyo, Japan) with an area of 10 mm × 10 mm was pressed to the targeted skin and then stripped. The SC-stripped tape was then cut into two equal parts. One half of the tape was used to prepare extracted lipid samples. The tape stripped was immersed into methanol with sonication for 10 min. The lipid extracts were dried using a nitrogen stream and were then dissolved in chloroform/methanol (99.5:0.5, by vol). This lipid solution was applied to a Sep-Pak Vac RC silica cartridge (Waters, Milford, MA) that had been conditioned by chloroform/methanol (99.5:0.5, by vol), followed by solid phase extraction with 10 ml chloroform/methanol (99.5:0.5, by vol) and 10 ml chloroform/methanol (99:5, by vol). The latter fraction, which contains the CERs, was dried using a nitrogen stream and then dissolved in n-hexane/2-propanol/formic acid (95:5:0.1, by vol). An amount of lipid extracts from the tape was approximately 5 g. The other was immersed into 0.1 M sodium hydroxide 1% (w/v) sodium dodecyl sulfate aqueous solution and was then incubated at 60°C for 2 h to obtain soluble proteins. After incubation, the solution was neutralized with 2 M hydrochloric acid aqueous solution.

Determination of soluble proteins
Concentrations of soluble proteins were determined according to the method of Smith et al. ( 41 ). A bicinchoninic acid protein assay kit (Pierce, Rockford, IL) and a microplate reader (Bio-Rad Laboratories, Hercules, CA) were used together with a calibration curve established using BSA.

Analysis using NP-HPLC-ESI-MS
The instrumentation was the same used for the comprehensive profi ling of CERs in the SC ( 24 ). Thus, an Agilent 1100 Series LC/MSD SL (single quadrupole) system equipped with an ESI source, ChemStation software, an 1100 well plate autosampler (Agilent Technologies, Palo Alto, CA), and an Inertsil SIL 100A-3 1.5 mm i.d. × 150 mm column (GL Science, Tokyo, Japan) was used. In this LC/MSD system, two mobile phases (A, n-hexane/2-propanol/formic acid, 95:5:0.1, by vol; B, n-hexane/2-propanol/50 mM ammonium formate aqueous solution, 25:65:10, by vol) were used with a fl ow rate of 0.1 ml/min. The mobile phases were consecutively programmed as follows: a linear gradient of A (100-90%) and B (0-10%) between 0 and 3 min; a linear gradient of A (90-0%) and B (10-100%) between 3 and 35 min; an isocratic elution of A (0%) and B (100%) for 5 min; a linear gradient of A (0-100%) and B (100-0%) between 40 and 50 min; and an isocratic elution of A (100%) and B (0%) from 50 to 80 min for column equilibrium. The injection volume was 10 l. The column temperature was maintained at 40°C. To enhance ionization of the molecular-related ions, 2.5 mM ammonium formate 2-propanol/methanol (50:50, by vol) at a fl ow rate of 0.1 ml/min was added to the mobile phase at the postcolumn position. ESI-MS parameters were as follows: polarity, positive; heater temperature of nitrogen gas, 300°C; fl ow of heated dry nitrogen gas, 8.0 l/min; nebulizer gas pressure, 20 psi; capillary voltage, 3500 V; fragmenter voltage, 150 V; cycle time; 1.15 s/ cycle; dwell time for each ion, 7 ms. For SIM measurement, 182 eluted at 3 to 4.5 min under the NPLC conditions, as predominant molecular-ions [M + NH 4 ] + (data not shown), whereas the shortest retention time of CER species is 5.5 min (C54 CER [NDS]). This indicates that SC lipids, including diacylglycerides, do not interfere with the detection of CERs in the SC as long as these NPLC conditions are used. Figure 1 shows representative multi-SIM chromatograms of CERs from the inner forearm and cheek SC under these NPLC-ESI-MS conditions. In this simultaneous SIM measurement for 182 kinds of ions, the S/N was about 30% of the SIM measurement for only one ion (data not shown), which indicates that the sensitivity in this simultaneous SIM measurement is lower than the ideal SIM measurement. Nevertheless, this method had more than a 10-fold higher sensitivity than the profi ling method using scan measurement ( 24 ), as the simultaneous SIM measurement using only one piece of tape with an area 10 mm × 10 mm had almost the same S/N of peaks as the scan measurement using three pieces of tape with areas of 18 mm × 25 mm. At the very least, this indicates that the simultaneous SIM measurement of CERs in the SC is much more sensitive than our previous scan measurement method for comprehensive profi ling.
To demonstrate the high sensitivity and quantitative performance of this method, LOD, LOQ, and calibration linearity for the 11 authentic different class CERs and the internal standard CER[N(17)S(18)] were examined. The LOD and LOQ for 10 l injection volumes of the standard solution were in the range 3.8-9.8 and 13-32 fmol, respectively ( Table 1 ). All correlation coeffi cients tested were greater than 0.9985 in the range 40-1280 fmol ( Table 1 ), although there were slightly higher values of the intercepts, which may be ascribed to the asymmetrical shape and slight tailing of the peaks. Table 1 confi rmed that the simultaneous SIM measurement has calibration linearity with a dynamic range in two fi gures and high sensitivity such that the LOD is at the fmol range. Based on these results, we determined that the internal standard should be added to each sample at a fi nal concentration of 50 nM (500 fmol in 10 l) so that its concentration corresponds to the center of the range of the calibration curves.

Procedure for estimating quantitative levels of overall CER species
We then examined the possibility of a procedure to estimate quantitative levels of overall CER species to comprehensively determine not only CERs available as authentic materials but also CERs that are not available. For this, relative responses were calculated for intra-day and for interday analyses. First, the 11 authentic different class CERs were used (  (20)] as the maximum in the intra-day analyses when the internal standard CER[N(17)S (18)] was regarded as being 1.00. The responses were relatively reproducible in the intra-day analyses as seen in the range of relative standard deviation (RSD%) of 1.3-8.3%. As for the responses of inter-day Determination of CER classes using TLC TLC was performed according to the method of Imokawa et al. ( 42 ). Briefl y, HPTLC-plates (silica gel 60 with concentrating zone, 20 cm × 10 cm; Merck, Darmstadt, Germany) were used for the separation. After aliquots of concentrated SC lipid samples collected from the inner forearm SC of two volunteers (samples A and B) were applied on a plate, it was developed twice with chloroform/methanol/acetic acid (190:9:1, by vol). After drying, the plates were sprayed with a 10% copper sulfate and 8% phosphoric acid solution and were then charred by heating at 180°C for 5 min. Each spot for CER classes was densitometrically determined using a TLC Scanner-3 photodensitometer (Camag, Muttenz, Switzerland) and a calibrated curve was established for each authentic material belonging to the same class as the targeted CERs. Their levels were compared with the total sum of the quantitative levels obtained by NPLC-ESI-MS.

Simultaneous SIM measurement in NPLC-ESI-MS
To detect CERs in the SC both with high sensitivity and with quantitative capability under the NPLC-ESI-MS conditions already optimized for the comprehensive profi ling of CERs ( 24 ), we examined the practical availability of simultaneous SIM measurement in NPLC-ESI-MS. As it was theoretically impossible to separately detect each of the overall species as molecular-related ions in the SIM measurement due to the presence of many isobaric species, we decided to detect CERs consisting of isobaric species as apparent peaks in the SIM measurement. In the mass spectrometer used, four channels can be simultaneously monitored, in each of which the simultaneous monitoring for 30 kinds of ions is possible. Thus, 120 ions are usually maximal in simultaneous SIM measurements. However, we considered that more than 120 kinds of ions could be detected in one analytical run if the kinds of monitoring ions were changed depending on detection time, because each CER class has a different retention time under the conditions of this NPLC separation ( 24 ). As a result, we designed a time program using the four channels (see supplementary Table I), where as many as 182 kinds of molecularrelated ions, including those derived from the overall species and the internal standard CER[N(17)S(18)], can be monitored.
To check that the NPLC-ESI-MS conditions are specifi c for CERs in SC, we examined the possibilities that SC lipids might ovelap with CERs. There are cholesterol, free fatty acids, cholesterol esters, and cholesterol sulfate within intercellular spaces in human SC ( 27,43,44 ), in addition to squalene, wax esters, and triglycerides that are excreted from sebaceous glands, and diglycerides, monoglycerides, and free fatty acids that are produced by their hydrolysis ( 45 ). Considering their molecular weights and polarities, it was readily judged that those lipids except for diacylglycerides do not overlap with the elution of any CERs under these NPLC conditions. As we could not eliminate a possibility that diacylglycerides might overlap with CERs, 10 kinds of authentic diacylglycerides were tested under these analytical conditions. As a result, all diacylglycerides tested that the inter-day analyses seem to be slightly less reproducible than the intra-day analyses for different class     Relative response values are a ratio of the slope of the calibration curve (100-1000 fmol) each for 11 authentic CERs to that of the internal standard CER Taken together with the above results, we designed a procedure for estimating quantitative levels of overall CERs in the SC using the simultaneous SIM measurement, based on the following assumptions: (1) relations between relative responses and total carbons in the linear regression observed in the 12 authentic CER[NS] may be fulfi lled in others; (2) relations between relative responses and hydrophilic structures in the results on the 11 authentic different class CERs may be fulfi lled in others; and (3) responses of isobaric species may be substantially similar to each other. In calculating levels of CERs according to those assumptions, we decided to use the responses obtained in the inter-day analyses since our analytical runs for the quantifi cation would be done day by day. The level of a CER species (QL, ng/ g protein) can be calculated according to equation 1 .

-5 X QL = 5 10 RA MW / R / PL
where the constant 5 × 10 Ϫ 5 is the molar quantity of the added internal standard (nmol); RA represents the area ratio of the targeted CER to the internal standard; MW represents the molecular weight of the targeted CER; R X represents the relative response of the targeted CER, which can be calculated according to equation 2 ; and PL represents the determined level of soluble proteins ( g).

X T C 4 6 N S R = R R / 1.19
where R TC46 represents a relative response of an authentic CER belonging to the same class as the targeted CER but having 46 total carbons as shown in Table 2 ; R NS represents the relative response of a CER[NS] with the same total carbons as the targeted CER, which can be calculated according to equation 3 ; and the constant 1.19 represents a relative response of CER[N(26)S (20)] in the inter-day analyses as shown in Table 3 . 1.0-4.4% and 1.2-3.9% in the intra-day and inter-day analyses, respectively. This indicates that the intra-day and inter-day analyses may be reproducible for CER species belonging to the same class, compared with those belonging to different classes. As seen in Table 3 , the responses seemed to increase with the increasing total carbons of the authentic CERs. Therefore we depicted the relationship between the total carbons and the responses for the 12 authentic CERs based on the inter-day result, where the regression curve is (relative response) = 0.0209 × (total carbon) + 0.2830 and the correlation coeffi cient (R) is 0.9320 ( Fig. 2 ). There was no obvious tendency on which of carbons of fatty acids or sphingoids mainly determines the responses, because the order of the responses among isobaric species was unrelated to the numbers of their carbons. Although not shown in a fi gure, the intra-day relationship was similar ( R = 0.9470). This was evidence indicating that the responses are positively correlated with the total carbons and implying that any relative responses where the constants 0.0209 and 0.2830 represent slope and intercept, respectively, of the regression line shown in Fig. 2 , and TC represents the number of total carbons of the targeted CER.

Recovery % of spiked authentic materials and comparison with the TLC method
We estimated the validity of comprehensive quantifi cation using the simultaneous SIM measurement in NPLC-ESI-MS. First, recovery percentage was obtained by comparing the determined levels for the 11 authentic different class CERs in spiked crude SC lipids with those in the nonspiked crude lipids. The crude lipids were extracted from a 10 mm × 10 mm piece of a SC-stripped tape. Table 4 shows the average and RSD of the recovery percentage, which are in the range of 81.8-94.9% and 2.1-12.5%, respectively. Although the RSD of CER[N (26) H (20)] and CER[A(26)P (20)] are somewhat higher, those are acceptable considering the practical analyses. This indicates that our method has acceptable accuracy and precision to determine the levels of the 11 CER species, even for the practical analysis in SC specimens of which some matrices coexist.
Quantitative levels of CERs determined in the SC using the newly developed method were compared with those using the conventional TLC method. Thus, crude SC lipid samples (samples A and B) were extracted from 10 consecutive SC-stripped tapes with areas of 15 mm × 15 mm from the inner forearm and were then applied to TLC analysis, followed by NPLC-ESI-MS analysis after the dilution of the lipid solutions. Since CER[NDS] and CER [ADS] are not separated from CER[NS] and CER[AS], respectively, on the TLC plate, those are presented as total sums in Table 5 . In the method using NPLC-ESI-MS, the individual CER species were quantitatively determined both by the simultaneous SIM measurement and by the procedure for estimating the quantitative levels. The summed levels of the different CER classes are shown in Table 5 . There were relatively greater differences in the levels between the TLC and NPLC-ESI-MS analyses, especially in the levels of the sum of CER[NDS] and CER [NS], and the sum of CER[ADS] and CER[AS] in sample A. Those differences were supposed to originate from vagueness of the TLC analyses for practical SC lipid samples. Thus, TLC spots of authentic CERs were defi nite due to application of only a single species in each spot, whereas the spots containing many CER species in the samples were so indefi nite that it was diffi cult to distinguish from the neighbors' spots. However, statistical analyses using Student's t -test showed no signifi cant differences between them ( P > 0.05 in all pairs). This indicates that the levels estimated using this method are not substantially different from those obtained by the conventional TLC analytical method.

Application
In the multi-SIM chromatograms of CERs from forearm and cheek SC obtained from a 35-year-old male, we de-  gram for SIM fi t to the chromatographic elution of 11 different CER classes, we could optimize novel conditions for simultaneously measuring as many as 182 kinds of molecular-related ions derived from the diverse CER species in the SC. The results on the 11 authentic different-class CERs indicate that this method is useful for highly sensitive quantifi cation. Thus, the calibration curves are linear over the dynamic range of at least two fi gures, and the LOD is in the fmol range, even during the simultaneous measurement of as many as 182 kinds of ions. Therefore, the use of SIM measurements to achieve highly sensitive quantifi cation has been very successful. As exemplifi ed in this study, the method is so highly sensitive that we can analyze diverse CER species in only one SC-stripped tape piece as small as 5 mm × 10 mm. Further, using this method, the authentic CERs are quantitatively recovered as seen in the recovery test in which the authentic CERs were spiked in the crude extracts of SC lipids, indicating that there are no ion suppression effects that interfere with the quantitative detection of the authentic CERs.
To comprehensively quantify diverse CER species in the SC, it would be ideal to get all authentic materials for individual CER species that exist in the SC and to produce their calibration curves in every analytical run together with the internal standard. However, such an analytical run would be realistically impossible. Referring to previous studies on triglycerides ( 3 ) and phospholipids ( 4 ), we used another strategy: (1) obtain a systematic rule for relative responses of CER species to the internal standard by using authentic CERs limited to but covering all CER that the method is highly sensitive for quantifying overall CER species in the SC, as they can be determined in only one SC-stripped tape piece as small as 5 mm × 10 mm.

DISCUSSION
This study was aimed at developing a method for quantifying overall CER species in the SC with high sensitivity based on an NPLC-ESI-MS technique already constructed for the comprehensive profi ling of CERs using scan measurement ( 24 ). As a result, by designing a new time pro-    (Table 2) can be also used for factor 2. Regarding factor 3, the average RSD of the slopes and intercepts in six regression curves, which were produced on six separate days ( between the forearm and cheek SC, because the levels differ from each other by 20%. However, we must wait for further analyses using multiple subjects to make it clear whether the differences in CER classes/species compositions between distinct SC regions are common in humans. Although the accuracy of this method was evaluated for the 11 authentic CERs by spiking them into crude SC lipids, we could not directly validate the quantitative recovery percentage for other CERs that are not available as authentic CERs but are present in the SC. Further, we could not validate the apparent quantitative levels for apparent peaks consisting of different isobaric species. However, we currently consider that quantitative levels estimated for nonvalidated CERs may roughly have similar accuracy to those for the authentic CERs. That's because the summed quantitative levels of CERs using the NPLC-ESI-MS technique were comparable to those determined by TLC, even though our method involves the theoretical procedure already described. Strictly speaking, those levels do not completely coincide, but such differences in levels may come from great differences in the analytical principles/procedures between NPLC-ESI-MS and TLC, especially from vagueness of the TLC analyses in practical SC lipid samples. Therefore, the conventional method using TLC that has been traditionally used to quantify the CER class levels in the SC can be replaced by this newer NPLC-ESI-MS species present in the SC and (2) theoretically estimate quantitative levels of CER species not available as authentic materials according to the systematic rule. To derive such a rule, we considered that at least two types of authentic CERs would be required: one type including CER standards belonging to 11 different classes but having identical total carbons, and the other including CER standards belonging to the identical class but having different total carbons. The fi rst type was determined to be 8 authentic CERs having 46 total carbons that consist of 26 and 20 carbons in the fatty acid and sphingoid moieties, respectively, for N-or A-containing CERs, and 3 authentic CERs having 68 total carbons consisting of 18, 30, and 20 carbons in E, O and sphingoid moieties, respectively, for EO-containing CERs, because those are commonly present in the SC ( 24 ). As for the second type, we prepared 12 authentic CERs belonging to CER[NS] with 30-50 total carbons. To evaluate differences in relative responses among isobaric species, the combination of constitutional fatty acids and sphingoids was also taken into consideration. Using the responses of those authentic CERs, a proper procedure for estimating the quantitative levels was developed.
One must bear in mind a total systematic error originating from this estimation when using it routinely as a comprehensive quantifi cation. According to equations 1, 2, and 3, when an error in measuring soluble proteins is ignored, the systematic error comes from the four following factors: (1) the area ratio of the targeted CER species to the internal standard; (2) the response of an authentic CER species belonging to the same class as the targeted species but having 46 total carbons; (3) the slope and intercept in the regression curve produced by different CER[NS] species; and (4) the response of CER[N(26) S (20)]. In addition, we have to consider a fi fth factor: the responses of different isobaric species. In the inter-day analyses in a sequential run, the average RSD of area ratios for the 11 authentic standards, 1.4% (data not shown) can be used for factor 1. The average RSD of the responses for closely related to CER functions in SC. It seems that the species with longer chains in fatty acid and/or sphingoid moieties are critically important for the understanding of the CER functions. The structures of those species were already identifi ed using NPLC-ESI-MS/MS ( 24 ). Based on this quantitative method, we can know their levels. However, when extremely higher or lower levels of the species with longer chains are observed in some SC samples, we should identify their structures in the samples using NPLC-ESI-MS/MS again because it might lead to new fi ndings on CER functions in the SC. As this method has extremely high sensitivity, we can quantify the overall species in only one SC-stripped tape piece as small as 5 mm × 10 mm, which allows an almost noninvasive analysis of CER species in human SC without any distress for the subjects tested. This implies that atopic dry skin and facial skin could be analyzed with this method, specimens of which are usually too painful to collect with organic solvent extractions ( 37 ) or cyano acrylate-stripping ( 42 ). This method enabling quantitative lipidomics approaches will be helpful for future studies clarifying the physicochemical and physiological roles of CER species in the SC, including advances in dermatological and cosmetic science. method that has advantages for highly sensitive quantifi cation of the overall CER species.
Herein we compare this novel method with previously reported methods for quantifi cation of CERs. As already noted, the LOQ for the 11 authentic CERs was 13-32 fmol using the simultaneous SIM measurement. Of all previous reports describing methods for the quantifi cation of CERs, the methods reported by Han ( 15 ) and by Masukawa and Tsujimura ( 46 ) are highly sensitive. In Han's method ( 15 ), which uses direct MS and neutral loss ion scan, the LOD was approximately 10 fmol (consumed) by examination using an authentic CER[NS] commercially available. Provided that the LOD and LOQ are determined according to the S/N of 3 and 10, respectively, our method would have almost the same LOQ as that method. The other method ( 46 ), which uses RPLC-ESI-MS, has an LOQ of 0.48-1.37 fmol for 13 authentic CER[NDS], CER[NS], CER[ADS] and CER[AS] that exist in human hair, which indicates higher sensitivity of that method than our method. However, it should be noted that in both those methods, the targeted species are restricted to 2 or 4 classes. At least our method covering as many as 11 classes has the highest sensitivity for CER species in the SC. On the other hand, regarding the comprehensive performance, our method should be compared with methods reported by Farwanah et al. (36)(37)(38)(39) and by Muñoz-Garcia et al. ( 47 ). Farwanah et al. (36)(37)(38)(39) developed a NPLC-APCI-MS method for analyzing CER species in the SC, which is superior to other earlier methods in terms of profi ling performance and has been used for clarifying differences in CER species between healthy and unhealthy human skin (36)(37)(38). However, the values derived from that method are not guaranteed as quantitative levels, so they eventually used both the levels of the CER class derived from TLC and the relative abundances of the CER species obtained from NPLC-APCI-MS for the comparison between healthy and unhealthy SC ( 37 ). This indicates that the method reported by Farwanah et al. (36)(37)(38)(39) is not very useful for the quantifi cation of overall CER species in the SC. In another method which used RPLC-atmospheric pressure photospray ionization-MS to analyze CER species in sparrow SC, reported by Muñoz-Garcia et al. ( 47 ) (18)] based on the assumption that the internal standard may have the same molar response as each of the CER species targeted. Although the total level of CERs obtained by that method agreed well with those determined by the TLC method, this agreement does not guarantee validity of the method for the quantifi cation of individual CER species. The authors actually noted in their paper that "until further verifi cation, our quantifi cation results remain inconclusive" ( 47 ). From the viewpoints of high sensitivity and quantitative performance, we can conclude that our novel method is superior to the other previous methods for analyzing CER species in the SC.
Finally, we describe perspectives using the newly developed method. In this study, we developed a new tool to detect subtle differences and changes which might be